基于图像局域特性的小波收缩去噪算法

传统的小波收缩去噪算法采用单一的阈值，它没有考虑到小波系数的类聚性，图像中重要小波系数类聚的局部具有重要的奇异特性，应降低阈值以保留图像的边缘；反之含有不重要小波系数的局部应提高阈值以消除更多的噪声，因此该文提出了一种基于图像局域特性的小波收缩自适应阈值去噪算法，这种算法根据图像局部的奇异性大小，选择适当的阈值进行去噪。实验结果表明，相对于传统的单一阈值去噪算法来说，新的算法可使滤波后图像的峰值信噪比有所提高，在一定程度上克服了单一阈值去噪算法无法滤除高质量图像中噪声的缺陷。

A Wavelet Shrinkage Denoising Algorithm Based on Local Characteristic of Image

The traditional wavelet shrinkage denoising algorithm adopts a single threshold,but it doesn't consider the concentricity of the wavelet's coefficients.The part of image where significant wavelet's coefficients congregate has significant singularity,thus we should decrease the threshold in order to protect image's edge.Otherwise,we should increase threshold for denoising.In this paper,a wavelet shrinkage denoising algorithm with adaptive thresholds based on local characteristic of image is proposed.The algorithm chooses appropriate thresholds for denoising according to the image's singularity.The experiment results demonstrate that the algorithm can improve the PSNR of the denoised picture compared with the traditional wavelet shrinkage denoising algorithm,and overcome the defect of the traditional wavelet shrinkage denoising algorithm that can't denoise the image of high-quality to some extent.